agentleFS
jonathanmalkinjonathanmalkin/jules/.claude/skills/build/SKILL.md

build

Skill65 starsChanged 7 months agoMITView on GitHub

Software development end-to-end: scope -> design -> plan -> execute -> deploy. Merges scope, writing-plans, executing-plans, and subagent-driven-development. Use when the user says 'build', 'implement', 'code', 'write a script', 'create a feature', or when /think chains here after identifying a concrete software deliverable.

  • Commits and pushes

169 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use backprop, finances and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • JuliusBrusseeSkill

    backprop

    Bug → spec protocol. When a bug is found or a test fails, trace the cause, decide whether a new §V invariant would catch recurrence, append to §B. This is the one non-obvious thing SDD does that plan-then-execute doesn't. Triggers on test failure, bug report, post-mortem, or explicit user ask.

    Also in 4 projects that use this

  • GerardoRdz96Skill

    finances

    Use when Gera talks about his money — logging an expense ("I spent X on Y"), checking where he stands this pay-period, closing a period, planning debt payoff, onboarding his financial picture, or asking for budgeting/finance opinions. Servy's financial brain. Triggers — "I spent", "log this expense", "where am I this period", "close the period", "how's my budget", "debt plan", "should I pay off", "finances onboard", "/finances".

    Also in 3 projects that use this

  • cpiprintSkill

    favicon

    Generate favicons from a source image

    Also in 3 projects that use this

  • madeehahassanSkill

    Go-Live

    Pre-launch checklist. Use before making any app public, sharing a link externally, or announcing a product launch. Adapts to Claude Code, claude.ai, and Cowork.

    Also in 3 projects that use this

  • TrevokeSkill

    define

    Use when a new feature request, bug report, or work request needs requirements defined before implementation

    Also in 3 projects that use this

  • HAR5HA-7663Skill

    shape

    Plan the UX and UI for a feature before writing code. Runs a structured discovery interview, then produces a design brief that guides implementation. Use during the planning phase to establish design direction, constraints, and strategy before any code is written.

    Also in 6 projects that use this

The file

We split this skill into the rules it gives an agent

11 rules. 0 are common, 10 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 155 edited versions we read keep these.

  • 10

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If standalone invocation: Run a lightweight scope pass.

    line 21
  2. Distinctive:

    After each batch: present what was done, what changed, any issues encountered

    line 94
  3. Distinctive:

    Wait for approval before proceeding to the next batch

    line 95
  4. Distinctive:

    If a task fails: report the error, retry once if transient, stop and report if persistent.

    line 96
  5. Distinctive:

    Only when [Your Name] explicitly says "just run it", "autonomous mode", or "don't wait for me."

    line 100
  6. Distinctive:

    Run tests after implementation changes.

    line 111
  7. Distinctive:

    Look up the project-specific deploy process:

    line 118
  8. Distinctive:

    Verify after deploy: Check the live URL or service endpoint.

    line 126
  9. Distinctive:

    If no deploy is needed (library code, internal tools), skip this step.

    line 128
  10. Distinctive:

    Write a descriptive commit message

    line 135
  11. Very specific:

    If the session is substantial, remind about /wrap-up

    line 137

“Common” is measured: kept by at least 6 in 10 of the 155 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Build

Software development end-to-end: scope -> design -> plan -> execute -> deploy -> ship. Handles the full lifecycle from "I want to build X" to "it's live and committed."

Arrives from: Direct invocation, or chained from /think when a concrete software deliverable is identified.


Step 0: Scope Verification

If arriving from /think: Requirements already exist (framing statement, constraints, minimum version). Confirm they're sufficient, then skip to Step 2.

If standalone invocation: Run a lightweight scope pass.

  1. What exactly are we building?
  2. Who uses it? (user, automated system, Jules)
  3. What's explicitly out of scope?
  4. What's the minimum version that delivers value?

Confirm scope before proceeding. One question at a time, with proposed answers. Don't frontload all questions -- ask, get the answer, then ask the next one.


Step 1: Existing Code Search

Dispatch a parallel Haiku Explore subagent to search for:

  • Related code, existing utilities, patterns to reuse
  • Project-level CLAUDE.md and config
  • Similar implementations in the codebase

Present findings in 3-5 bullets. Don't re-research what was already found in /think.

Reuse over rebuild. Actively search for existing functions and utilities that can be reused. Never propose new code when a suitable implementation already exists in the codebase.


Step 2: Design

Architecture sketch: 3-5 bullets covering components, data flow, interfaces.

Identify dependencies, risks, and open questions. If major design decisions are needed, surface them as Decision Cards:

[DECISION] Summary | Rec: X | Risk: Y | Reversible? Yes/No

One-way doors get explicit approval before proceeding.

Design review. Dispatch a fresh Sonnet subagent with the design sketch. Prompt: "Are we building the right thing? Simpler approach? Edge cases? Missing tests? Feasibility?"

Self-review (run internally, surface only issues):

  1. Failure post-mortem: "3 months from now, this failed. What happened?"
  2. Over-engineering post-mortem: "3 months from now, this was massive overkill. What didn't we need?"
  3. Unstated assumptions: What are we assuming without acknowledging?
  4. Open questions: What would a skeptical reviewer ask?

Step 3: Plan

Break work into batches of 3-7 tasks each. A batch is a logical unit of work that can be reviewed as a set.

Each task includes:

  • File: File to modify (or create)
  • Action: What to do (specific action, not vague direction)
  • Verify: How to confirm it worked (test command, expected output, visual check)

Task granularity: one action per 2-5 minutes (write test -> run it -> implement -> run again -> commit).

Save plan to both locations:

  1. ~/.claude/plans/YYYY-MM-DD-[description].md
  2. Documents/Field-Notes/Plans/YYYY-MM-DD-[description].md

Present plan for approval before executing.


Step 4: Execute

Two modes. Batch is default. Autonomous requires explicit opt-in.

Batch Mode (default)

  1. Execute one batch of tasks
  2. After each batch: present what was done, what changed, any issues encountered
  3. Wait for approval before proceeding to the next batch
  4. If a task fails: report the error, retry once if transient, stop and report if persistent. Don't silently swallow errors.

Autonomous Mode

Only when [Your Name] explicitly says "just run it", "autonomous mode", or "don't wait for me."

  1. Dispatch one subagent per independent task using isolation: "worktree" for filesystem safety
  2. Two-stage review per task:
    • Stage 1 (spec compliance): Does the output match the task spec?
    • Stage 2 (code quality): Is the code clean, tested, and consistent with project patterns?
  3. Continuous progress -- don't wait for human sign-off between tasks
  4. Report results when all tasks complete

Both modes

  • Run tests after implementation changes. Don't leave broken code.
  • Subagents cannot write to subdirectory repos (Code/jules/, Code/kink-archetypes/). Main session handles those writes.

Step 5: Deploy

Look up the project-specific deploy process:

ProjectDeploy Command
[your-domain]Scripts/push-[your-handle].sh
Quiz appScripts/deploy-quiz.sh
OtherPresent the deploy command for review before running

Verify after deploy: Check the live URL or service endpoint. "Deploy succeeded" is not the same as "it works."

If no deploy is needed (library code, internal tools), skip this step.


Step 6: Ship

  1. Stage specific files for commit (never git add . or -A)
  2. Write a descriptive commit message
  3. git push
  4. If the session is substantial, remind about /wrap-up

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.